Inspiration

Online shopping and digital payments are convenient, but users increasingly face risks such as fake sellers, suspicious deals, phishing attempts, and payment scams. We wanted to build a solution that helps people evaluate purchase risks before making payment decisions. This inspired us to create PayShield AI, an AI-powered purchase risk intelligence platform.

What it does

PayShield AI analyzes purchase details and identifies potential warning signs using AI-powered risk assessment and rule-based fraud detection.

Key features include:

AI Fraud Detection:** Evaluates purchase details and generates a risk score and risk level. Transaction History Analysis:** Uses previous transaction patterns to identify unusual activity. Live Security Research:** Retrieves relevant security guidance from trusted sources using Tavily. Security Alert Center:** Displays purchase-related warnings and recommendations. Smart Spending Analytics:** Helps users understand their transaction activity and spending patterns. Payment Integration:** Supports PayPal Sandbox and Razorpay Test Mode for testing payment workflows. Secure Authentication:** Uses password hashing and JWT-based authentication with HTTP-only cookies.

How we built it

I developed PayShield AI using a full-stack architecture. Frontend: React.js and Vite provide an interactive dashboard for purchase analysis, transactions, alerts, and analytics. Backend: Node.js and Express.js handle authentication, payment workflows, AI analysis, and API requests. Database: MongoDB stores user accounts, transactions, and analysis records. AI: Groq API powers AI-based purchase risk assessment and recommendations. Live Research: Tavily Search API retrieves relevant security information from selected trusted domains. Payments: PayPal Sandbox and Razorpay Test Mode support payment integration and verification workflows.

Challenges we ran into

One major challenge was combining AI-generated risk assessments with rule-based fraud detection while keeping the results understandable to users.

I also worked through payment verification, secure authentication, MongoDB integration, API communication, and frontend-backend integration.

Another challenge was integrating live web research while ensuring that search results were treated as reference information rather than proof of fraud.

Accomplishments that we're proud of

  • Built a working full-stack AI-powered purchase risk analysis platform.
  • Integrated AI-based risk scoring and rule-based fraud indicators.
  • Implemented live security research with source links.
  • Developed transaction history and security alerts.
  • Integrated PayPal Sandbox and Razorpay Test Mode.
  • Created a unified dashboard with spending analytics.

What we learned

I gained practical experience in full-stack development, REST APIs, MongoDB, authentication, payment integration, AI API integration, live web research, and building a complete application.

I also learned how to combine multiple technologies into a single application and handle security-related workflows.

What's next for PayShield AI

I plan to improve fraud detection using more advanced behavioral analysis, expand trusted security research sources, introduce downloadable risk reports, improve anomaly detection, and explore real-time transaction monitoring.

Our long-term vision is to make online purchasing safer by giving users clearer, more actionable security insights before they pay.

Built With

Share this project:

Updates

Submission history